The Agent Skills Directory

[SAFE]: The skill provides documentation and examples for 'dlt' (Data Load Tool), an open-source Python library. All code samples demonstrate legitimate data engineering tasks such as loading data from GitHub and HubSpot APIs into data warehouses (BigQuery, DuckDB).
[CREDENTIALS_UNSAFE]: No hardcoded credentials were found. The skill correctly demonstrates best practices by using dlt.secrets to access sensitive tokens, which are expected to be provided via environment variables or secure configuration files rather than hardcoded in the script.
[EXTERNAL_DOWNLOADS]: The skill references downloads from well-known and trusted sources. It provides instructions to install the 'dlt' package from the official Python Package Index (PyPI) and makes HTTP requests to established API endpoints (GitHub, HubSpot). These represent standard, safe library functionality.
[DATA_EXFILTRATION]: No data exfiltration patterns were detected. The data movement demonstrated in the examples is the primary purpose of the skill (ETL/Data Ingestion) and targets legitimate destinations specified by the user.
[INDIRECT_PROMPT_INJECTION]: While the skill is designed to process external API data, it includes examples of 'Data Contracts' which enforce schema validation, acting as a security and integrity control for ingested content. The attack surface for prompt injection is minimal as the data is handled as structured records for database insertion.

dlt